Chapter 2–4: Data Description, Statistics, Ethics, Replication, Epigenetics, and Neuroimaging

Describing Data, Sample Size, and Descriptive Statistics

• Bigger sample size generally improves reliability and the generalizability of top-line results, but practical/usability and cost limitations can cap how large samples can be.
• At a high level, we describe data before diving into analysis; formal stats aren’t the focus here.
• Measures of central tendency:
  • Mean $\bar{x} = \dfrac{1}{n} \sum<em>{i=1}^n x</em>i$
  • Median: the middle value; for even n, $\text{Median} = \dfrac{x<em>{(\tfrac{n}{2})} + x</em>{(\tfrac{n}{2}+1)}}{2}$
  • Mode: the most frequently occurring value
• Utility of mean, median, and mode depends on the data and research question; researchers may use more than one metric depending on the context.
• Outliers can skew the mean; example with a feeling thermometer (0 to 100): most responses cluster high (80–100) but a few respond with 0; including zeros lowers the mean, excluding them raises it. This illustrates why outliers deserve attention and why researchers check for them.
• Outliers can be leveraged or manipulated to fit an argument if not predefined (e.g., deciding post hoc to exclude certain scores without preregistration). Always check how outliers were defined and handled in a study.
• Practical example: housing prices in Northwest Arkansas can skew the mean due to very expensive homes in Bentonville; the median may better reflect a typical value in such a skewed distribution.
• Other descriptive measures of variability:
  • Range: $\text{Range} = \max(x<em>i) - \min(x</em>i)$
  • Standard deviation: $s = \sqrt{\dfrac{1}{n-1} \sum<em>{i=1}^n (x</em>i - \bar{x})^2}$
  • Variance: $\text{Var}(X) = \dfrac{1}{n-1} \sum<em>{i=1}^n (x</em>i - \bar{x})^2 = s^2$
  • Invariance (conceptual): variance is the average of squared deviations from the mean; lower variability means tighter clustering around the mean
• How to describe variability depends on the context; sometimes range or standard deviation is more informative than simply the mean
• Descriptive stats set up the data for inferential stats; the choice of metric affects interpretation

Statistics, Inference, and the 5% Rule

• Inferential goal: determine if observed differences are statistically meaningful rather than due to chance.
• Common framework: p-values and statistical significance.
• The 5% rule: a common conventional cutoff is p < 0.05, meaning there is less than a 5% chance that the observed difference (or a more extreme one) would occur if the null hypothesis were true.
• A p-value is the probability, under the null hypothesis, of obtaining data as extreme as or more extreme than what was observed.
• Interpretive nuance:
  • A p-value below 0.05 does not guarantee practical significance or importance; it only speaks to statistical significance.
  • A result can be statistically significant but not practically meaningful in real-world terms (and vice versa).
• Researchers sometimes adopt more stringent thresholds to improve replicability, e.g., p < 0.01 or p < 0.001; these reductions make it harder to claim significance and emphasize robustness.
• Important caveat: statistical significance is an arbitrary cutoff; results should be interpreted in light of effect sizes, confidence intervals, study design, and prior evidence.
• Replicability concerns have motivated discussions about lowering thresholds or adopting preregistration and transparency practices to curb questionable research practices.

Ethics, IRB, Informed Consent, Deception, and Data Protection

• Researchers are ethically and legally obligated to conduct studies in an ethical manner.
• Institutional Review Board (IRB): a committee at universities that reviews studies involving human subjects to ensure they are ethical and minimize risk.
• Informed consent: participants are informed about the study, its procedures, potential risks, and their right to withdraw; participation should be voluntary.
• Deception: may be used in some studies when full disclosure would bias results or when the study’s artificial lab setting requires it; debriefing is required afterward to explain the true purpose and address any lingering concerns.
  • Post-debriefing, participants may opt out and request that their data not be used.
• Data handling:
  • Anonymity: participants cannot be identified in any aggregated data—no link between data and identity.
  • Confidentiality: linking information may exist in data files, but identifying information is kept separate and protected.
  • Informed consent also covers how data will be stored, who will access it, and whether data will be shared.
• IRB composition includes professional staff and faculty from various disciplines; biases can influence decisions, which is acknowledged and mitigated by oversight and standards.
• Animal research is subject to ethical review as well; IRBs or equivalent committees oversee animal studies.
• Historical note: some past studies caused ethical concerns; IRBs aim to prevent repeats of unethical treatment.
• Techniques like fMRI and other neuroimaging have safety profiles; ongoing assessment is part of ethical practice.

Replication Crisis, Reproducibility, and Scientific Integrity

• Replication crisis: many classic psychology studies fail to replicate when the methods are repeated with new samples.
• Contributing factors discussed in class:
  • Low validity of core scales and measures (poor operationalization of constructs).
  • Measurement quality and how concepts are operationalized (e.g., depression scales vs single-item questions).
  • Small sample sizes in older studies (e.g., 20 participants) limiting generalizability.
  • Data fabrication or selective reporting (p-hacking, outlier exclusion post hoc).
  • Questionable research practices (flexible data analysis, selective reporting).
  • Publication bias and noise over time—contextual or societal changes can alter results.
• Replicability of gender bias in STEM across eras illustrates how replication attempts can yield different results due to time, populations, or methods changes.
• Replication does not automatically imply higher quality; critical analysis of replication design and context is essential.
• Promoting integrity and trust in science involves:
  • Adversarial collaboration to test competing theories openly.
  • Pre-registration of hypotheses, methods, and analysis plans (e.g., via OSF—Open Science Framework).
  • Pre-registering the entire experimental plan and planned analyses to reduce data-dignose and p-hacking.
  • Data sharing and making data available in online repositories when possible, while respecting confidentiality.
  • Acknowledging that some data cannot be shared due to privacy or ethical concerns.
• Practical takeaway: replication is essential but must be interpreted within context; it is a means to validate knowledge, not an automatic guarantee of truth.

Epigenetics, Nature–Nurture, and Gene–Environment Interaction

• Core idea: genes and environment interact; environment can influence gene expression (epigenetics).
• Classic rat study example: nurturing behavior in rat mothers (licking pups) affects offspring development and later behavior; cross-fostering studies show environmental context can shape outcomes even with the same genes.
• Mechanism: environmental exposure can alter gene expression (e.g., DNA methylation) without changing the DNA sequence; these changes can persist across development and may even be transmitted transgenerationally in some contexts.
• Key takeaway: genes set potentials, environment can modulate expression; sometimes environment leaves a lasting imprint on biology.
• Human epigenetics examples:
  • Great Depression-era cohorts show long-term effects on DNA and possibly transgenerational traits due to severe early-life environment.
  • Traits like freckles or obesity risk can be modulated by gene expression in response to environmental factors.
• Study designs in behavioral genetics:
  • Twin studies: identical genes, different environments help separate genetic and environmental contributions.
  • Adoption studies: separate upbringing environment from genetic background.
• Behavioral and clinical approaches:
  • Clinical case studies assess brain injuries or diseases to understand brain-behavior links.
  • Behavioral neuroscientists use experimental interventions (e.g., brain stimulation or lesions) to test causal roles of brain regions.
• Conceptual link to neuroscience: how environment can influence the brain at genetic and neural levels through epigenetic mechanisms.

Brain Imaging, Neurophysiology, and Measurement Tools

• A toolkit of methods to study the brain includes:
  • Electrodes and sensors (EEG-like setups) to measure electrical activity and arousal during tasks or sleep states.
  • PET scans (Positron Emission Tomography): track radioactive glucose or other tracers to visualize metabolic activity in the brain. Higher glucose uptake indicates higher metabolic demand in active areas; color maps reflect uptake levels.
  • fMRI (functional MRI): measures brain activity by detecting changes in blood flow (BOLD signal) related to neural activity; reflects the hemodynamic response rather than direct neural firing.
• How fMRI works (high-level):
  • Neurons firing increases local blood flow, changing oxygenated hemoglobin levels, which alters magnetic signals detectable by MRI.
  • The brain is imaged in slices (e.g., 20 horizontal slices) to build a 3D picture of activity over time.
  • Scans involve radiofrequency pulses to reorient hydrogen atoms; data are collected as a time series to infer activity patterns.
• Practical notes on imaging:
  • Mock scanners can acclimate participants (especially children) to reduce movement and anxiety.
  • Safety: long-standing use with no significant adverse effects reported; animal safety studies support safety in long-term exposure in research settings.
• Cautionary example: the thinking dead salmon study (circa 2009) highlighted that neuroimaging methods can produce false positives if statistical thresholds and data handling aren’t stringent enough; it spurred reforms to improve reliability and interpretation of brain-imaging results.
• Takeaway on neuroimaging:
  • No method is perfect; each has limitations and requires careful statistical and methodological safeguards.
  • Researchers must balance the allure of powerful imaging techniques with rigorous science and transparent reporting.

Practical and Theoretical Connections

• The chapter connects descriptive data description, inferential statistics, ethics, replication, and neuroimaging into a cohesive view of how psychology studies are designed and interpreted.
• Real-world implications:
  • Understanding data description helps prevent misinterpretation of results due to outliers or skewed distributions.
  • Ethical practices (IRB, informed consent, debriefing) protect participants and ensure credibility of research findings.
  • Replication and preregistration promote scientific integrity and public trust in findings.
  • Epigenetics and gene–environment interactions highlight the complexity of “nature vs. nurture,” reminding us that behavior emerges from dynamic interplays between biology and environment.
• Preview for next topics: nervous system explanations and additional neuroimaging techniques that will expand on how brain structure and function relate to behavior.

Key Formulas and Definitions (Quick Reference)

• Mean: $\bar{x} = \dfrac{1}{n} \sum<em>{i=1}^n x</em>i$
• Median: midpoint value; even n uses average of two central values
• Mode: most frequent value
• Range: $\text{Range} = \max(x<em>i) - \min(x</em>i)$
• Variance (sample): $\text{Var}(X) = \dfrac{1}{n-1} \sum<em>{i=1}^n (x</em>i - \bar{x})^2$
• Standard deviation (sample): $s = \sqrt{\dfrac{1}{n-1} \sum<em>{i=1}^n (x</em>i - \bar{x})^2}$
• p-value cutoff (typical): p < 0.05 (or stricter thresholds like p < 0.01 or p < 0.001)
• Conceptual note: Statistical significance ≠ Practical significance; both should be considered in interpretation

Connections to Prior and Real-World Relevance

• Links to foundational statistics (descriptive vs inferential) and research design principles (validity, reliability, sample size).
• Relates to ethics in research practices and compliance with IRB guidelines to ensure safe, fair, and credible science.
• Demonstrates how replication initiatives and open science practices (pre-registration, data sharing) contribute to more trustworthy science across disciplines.
• Illustrates how modern neuroscience combines multiple modalities (EEG, PET, fMRI) to understand brain–behavior relationships, while acknowledging limitations and the need for careful interpretation.